python - pyqt's signals vs directly calling a method -

i'm trying come grips why 1 want use pyqt signal/slot instead of calling class method directly.

here simple snippet of code illustrates how see it:

from pyqt4 import qtgui, qtcore  class simple_signal(qtcore.qobject):     single_signal = qtcore.pyqtsignal()     double_signal = qtcore.pyqtsignal()      def __init__(self):         super().__init__()         # connect 1 signal 1 slot         self.single_signal.connect(self.handle_signal_choice1)          # connect signal 2 slots         self.double_signal.connect(self.handle_signal_choice1)         self.double_signal.connect(self.handle_signal_choice2)          # connect 1 method         self.method = self.handle_signal_choice3      def handle_signal_choice1(self):            print("simple_signal - choice1 received")      def handle_signal_choice2(self):         print("simple_signal - choice2 received")      def handle_signal_choice3(self):         print("simple_signal - choice3 received")  if __name__ == "__main__":     simple_signal = simple_signal()     print('calling choice1 via single_signal')     simple_signal.single_signal.emit()      print('calling choice1 , choice2 via double_signal')     simple_signal.double_signal.emit()      print('calling choice3 via class method')     simple_signal.method() 

one immediate advantage signals seems be: it's easy attach signal 2 slots. however, method example without fuss.

the method example seems better cpu time results (ie faster) signal/slot example, advantage may lost if needed many slots called 1 signal.

in mind, calling method cleaner messing signal/slot setup.

either work in qwidgets.

do need more complex situation signals/slots better choice directly calling method?

thanks.

edit: being curious time consumed in each method, did timeit checks.

add these timeit lines code, , replace each 'print' line in original code (ie in handle_signal_choice1) 'pass' printing shell doesn't slow things down.

result=timeit.timeit('simple_signal.single_signal.emit()', 'from __main__ import simple_signal', number=10000) print('calling choice1 via single_signal') print('timeit says : ', result)  result=timeit.timeit('simple_signal.double_signal.emit()', 'from __main__ import simple_signal', number=10000) print('calling choice1 , choice2 via double_signal') print('timeit says : ', result)  result=timeit.timeit('simple_signal.method()', 'from __main__ import simple_signal', number=10000) print('calling choice3 via method') print('timeit says : ', result) 

this gives me following results, upholds suspicion single method call faster single signal/slot:

now start timeit tests calling choice1 via single_signal   timeit says :  0.010595089312688225 calling choice1 , choice2 via double_signal  timeit says :  0.014604222516910098 calling choice3 via method timeit says :  0.0016033988748859057 

that's quite big difference if drive socket event loop via signals/slots instead of methods. button clicking shouldn't matter.

the reason use signals , slots isn't speed, it's loose coupling. using signals , slots allows better reuse. there's nice paper goes on motivation , benefits. add indirection, it's worth it, imho.